fix(CEA): Fix multi byte language support in CEA-708 (#7837)

Fixes #7829

Author: avelad

lib/cea/cea708_service.js

@@ -62,7 +62,7 @@ shaka.cea.Cea708Service = class {
     // Control codes are in 1 of 4 logical groups:
     // CL (C0, C2), CR (C1, C3), GL (G0, G2), GR (G1, G2).
     if (controlCode >= 0x00 && controlCode <= 0x1f) {
-      return this.handleC0_(controlCode, pts);
+      return this.handleC0_(dtvccPacket, controlCode, pts);
     } else if (controlCode >= 0x80 && controlCode <= 0x9f) {
       return this.handleC1_(dtvccPacket, controlCode, pts);
     } else if (controlCode >= 0x1000 && controlCode <= 0x101f) {
@@ -155,17 +155,42 @@ shaka.cea.Cea708Service = class {
 
   /**
    * Handles C0 group data.
+   * @param {!shaka.cea.DtvccPacket} dtvccPacket
    * @param {number} controlCode
    * @param {number} pts
    * @return {?shaka.extern.ICaptionDecoder.ClosedCaption}
    * @private
    */
-  handleC0_(controlCode, pts) {
+  handleC0_(dtvccPacket, controlCode, pts) {
     // All these commands pertain to the current window, so ensure it exists.
     if (!this.currentWindow_) {
       return null;
     }
 
+    if (controlCode == 0x18) {
+      const firstByte = dtvccPacket.readByte().value;
+      const secondByte = dtvccPacket.readByte().value;
+
+      const isTextBlock = (b) => {
+        return (b >= 0x20 && b <= 0x7f) || (b >= 0xa0 && b <= 0xff);
+      };
+
+      if (isTextBlock(firstByte) && isTextBlock(secondByte)) {
+        const toHexString = (byteArray) => {
+          return byteArray.map((byte) => {
+            return ('0' + (byte & 0xFF).toString(16)).slice(-2);
+          }).join('');
+        };
+        const unicode = toHexString([firstByte, secondByte]);
+        // Takes a unicode hex string and creates a single character.
+        const char = String.fromCharCode(parseInt(unicode, 16));
+        this.currentWindow_.setCharacter(char);
+        return null;
+      } else {
+        dtvccPacket.rewind(2);
+      }
+    }
+
     const window = this.currentWindow_;
     let parsedClosedCaption = null;
 

lib/cea/dtvcc_packet_builder.js

@@ -161,6 +161,21 @@ shaka.cea.DtvccPacket = class {
     }
     this.pos_ += numBlocks;
   }
+
+  /**
+   * Rewinds the provided number of blocks in the buffer.
+   * @param {number} numBlocks
+   * @throws {!shaka.util.Error}
+   */
+  rewind(numBlocks) {
+    if (this.pos_ - numBlocks < 0) {
+      throw new shaka.util.Error(
+          shaka.util.Error.Severity.CRITICAL,
+          shaka.util.Error.Category.TEXT,
+          shaka.util.Error.Code.BUFFER_READ_OUT_OF_BOUNDS);
+    }
+    this.pos_ -= numBlocks;
+  }
 };
 
 /**

test/cea/cea708_service_unit.js

@@ -123,6 +123,35 @@ describe('Cea708Service', () => {
     expect(captions).toEqual(expectedCaptions);
   });
 
+
+  it('decodes multibyte unstyled caption text', () => {
+    const controlCodes = [
+      ...defineWindow,
+      // Series of C0 control codes that add multi-byte text.
+      0x18, 0xb9, 0xd9, 0x18, 0xb7, 0xce, // 맙, 럎
+    ];
+
+    const packet1 = createCea708PacketFromBytes(controlCodes, startTime);
+    const packet2 = createCea708PacketFromBytes(hideWindow, endTime);
+
+    const text = '맙럎';
+    const topLevelCue = CeaUtils.createWindowedCue(startTime, endTime, '',
+        serviceNumber, windowId, rowCount, colCount, anchorId);
+    topLevelCue.nestedCues = [
+      CeaUtils.createDefaultCue(startTime, endTime, /* payload= */ text),
+    ];
+
+    const expectedCaptions = [
+      {
+        stream,
+        cue: topLevelCue,
+      },
+    ];
+
+    const captions = getCaptionsFromPackets(service, packet1, packet2);
+    expect(captions).toEqual(expectedCaptions);
+  });
+
   it('setPenLocation sets the pen location correctly', () => {
     const controlCodes = [
       ...defineWindow,